ADJ-AVP S1 · Ground truth and calibration
Attribution versus patching, adjudicated against planted ground truth
Refuted K-avp fired
frozen 2026-07-18 · adjudicated 2026-07-19 · ev:a34457e83cd4eec10d080804a2533e82 · spec:8a5635a89a8d5
The frozen predictions
- H-avp patching's ranking matches the organism's marker-responsive components (the twin-activation-delta answer key) better than attribution's
recovery_gap > 0registered p 0.65 recovery_gap Value -0.5660377358490566 Subject ADJ-AVP Threshold > 0 Trust Adjudicated Evidenceev:a34457e83cd4eec10d080804a2533e82Refuted
K-avp fired on recovery_gap_ci_low <= 0
, observed -0.7736
: the two instruments are statistically tied against the marker-response key on the organism; neither localization can be preferred, undercutting every attribution-based number (a noisy key, visible in the emitted margin distribution, is the disclosed alternative reading). A fired kill criterion is a registered negative
result, not a failure of the machinery.
Registered companions
-
patching_recovery_aucpatching_recovery_auc Value 0.4339622641509434 Subject ADJ-AVP Trust Adjudicated Evidenceev:a34457e83cd4eec10d080804a2533e82 -
attribution_recovery_aucattribution_recovery_auc Subject ADJ-AVP Trust Adjudicated Evidenceev:a34457e83cd4eec10d080804a2533e82 -
real8b_avp_spearmanreal8b_avp_spearman Value -0.3735805860805861 Subject ADJ-AVP Trust Adjudicated Evidenceev:a34457e83cd4eec10d080804a2533e82 -
recovery_gap_ci_lowrecovery_gap_ci_low Value -0.7735849056603774 Subject ADJ-AVP Trust Adjudicated Evidenceev:a34457e83cd4eec10d080804a2533e82 -
recovery_gap_ci_highrecovery_gap_ci_high Value -0.4276729559748428 Subject ADJ-AVP Trust Adjudicated Evidenceev:a34457e83cd4eec10d080804a2533e82